Different Domain Interactions Are Involved in the Binding of Tissue

Department of Biochemistry, Queen Mary and Westfield College, University of London, Mile End Road, London El 4NS,U.K., and Department of Oncology ...
0 downloads 0 Views 3MB Size
Biochemistry 1994, 33, 2089-2095

2089

Different Domain Interactions Are Involved in the Binding of Tissue Inhibitors of Metalloproteinases to Stromelysin-1 and Gelatinase At Quang Nguyen,* Frances Willenbrock,S Mark I. Cockett,I Mark O'Shea,t Andrew J. P. Docherty,l and Gillian Murphy'.$ Department of Cell and Molecular Biology, Strangeways Research Laboratory, Worts Causeway, Cambridge CB1 4RN, U.K., Department of Biochemistry, Queen Mary and Westfield College, University of London, Mile End Road, London E l 4NS, U.K., and Department of Oncology Biology, Research Division, Celltech Limited, 216 Bath Road, Slough SLI 4EN, U.K. Received September 22, 1993; Revised Manuscript Received December 14, 1993" ABSTRACT: The matrix metalloproteinases gelatinase A and stromelysin- 1 have definable N-terminal

(catalytic) and C-terminal domains. In order to analyze their interactions with the N- and C-terminal domains of the tissue inhibitors of metalloproteinases TIMP-1 and -2, mutants of both the enzymes and the inhibitors were prepared in which the C-terminal domains had been deleted. Since the Ki values for T I M P inhibition of the matrix metalloproteinases are in the picomolar range, it was not possible to measure these accurately within the sensitivity of available activity assays. Rate constants for the association of the wild-type proteins were therefore determined and systematically compared with those for the deletion mutants. It was found that TIMP-1 binds more rapidly than TIMP-2 to stromelysin-1 and that the C-terminal domain of the enzyme does not affect the rate of association of enzyme and inhibitor. This is in contrast to gelatinase A, where the C-terminal domain has been shown to play an important role in increasing the rate of complex formation with the TIMPs (Willenbrock et al., 1993). The TIMPs are also comprised of an N - and C-terminal domain. By deletion mutagenesis, we found that the C-terminal domain of both TIMPs contributed less to the rate of complex formation with stromelysin-1 than to that with gelatinase A. Hybrids of the N- and C-terminal domains of gelatinase A and stromelysin-1 were prepared and used to analyze further the differences in domain interactions with the TIMPs. They demonstrated that the interactions between the C-terminal domains of enzyme and inhibitor can occur irrespective of the nature of the N-terminal domain. We can conclude that the TIMPs have two major binding regions which associate in different ways with the domains of the enzymes gelatinase A and stromelysin- 1. The N-terminal domains of the TIMPs bind to the enzyme catalytic domains to inhibit activity. The T I M P C-terminal domain acts to increase the association rate constant by binding to the N-terminal domain of stromelysin or the C-terminal domain of gelatinase A.

The matrix metalloproteinases(MMPs) are thought to play an important role in the physiological and pathological remodeling of connective tissues, where they are capable of hydrolyzing collagens, proteoglycans, and other matrix components. Extracellular activity of the MMPs is controlled in vivo by a family of inhibitors, the tissue inhibitors of MMPs, or theT1MPs.l Todate, the primarysequencesofthreehuman TIMPs have been obtained by cDNA cloning: TIMP-1 (Docherty et al., 1985), TIMP-2 (Boone et al., 1990), and TIMP-3 (Apteet al., 1993). Studies arenow directed toward the understanding of the mechanism of action of the TIMPs.

Early studies showed that they made tight-bindingnoncovalent complexes with active MMPs with a 1:l stoichiometry (Cawston et al., 1981, 1983; Welgus et al., 1985; Okada et al., 1986; Murphy et al., 1987). TIMP-1 and TIMP-2 have previously been shown to have similar abilities to inhibit all the active MMPs when assessed using macromolecular substrates (Ward et al., 1991). This is consistent with the fact that these inhibitors have related primary and secondary structures. The TIMPs consist of six disulfide-bonded loops, and deletion mutagenesis studies have demonstrated that two structurally distinct domains can be defined, the N-terminal domain consisting of loops 1-3 and the C-terminal domain consisting of loops 4-6 (Murphy et al., 1991; Willenbrock et t Q.N.is supported by the WellcomeTrust, U.K., F.W. by the Medical Research Council, U.K., and G.M. by the Arthritis and Rheumatism al., 1993). The N-terminal domain can act as a functional Council, U.K. inhibitor (Murphy et al., 1991; DeClerck et al., 1993) and is * To whom correspondence should be addressed. thought to bind at or near the enzyme active site, since * Strangeways Research Laboratory. 8 Queen Mary and Westfield College. competition has been observed between TIMP-1 and lowI Celltech Ltd. molecular-weight synthetic inhibitors that are directed at the Abstract published in Advance ACS Abstracts, February 1 , 1994. catalytic zinc (Lelikvre et al., 1990). Very tight complexes I Abbreviations: TIMP, tissue inhibitor of metalloproteinases;A~27-184 are formed between the TIMPs and MMPs with Ki values of TIMP-1 (N-TIMP-I), TIMP-1 from which amino acids 127-184 have been deleted; A128-194 TIMP-2 (N-TIMP-2),TIMP-2 from which amino less than 1 nM. We have previously reported approximate Ki acids 128-194 have been deleted; A248-4~)stromelysin (N-SL), prosvalues of 2 and 30 pM, respectively, for the inhibition of tromelysin from which amino acids 248-460 have been deleted; N-GL.Cgelatinase A and a mutant of gelatinase A that lacks the SL, hybrid of gelatinase A residues 1-417 and stromelysin-1 residues C-terminal domain ( 4 1 8 4 3 1 gelatinase A, N-GL) by either 248-460; N-SLC-GL, hybrid of stromelysin-1 residues 1-247 and gelatinase A residues 418-63 1; PCR, polymerase chain reaction; TIMP-1 or TIMP-2 (Murphy et al., 1992b). For theinhibition McaPLANvaDpaAR, (7-methoxycoumarin-4-y1)acetyl-Pro-Leu-Ala- of stromelysin and a similarly truncated mutant, Ki values are Norval- [3-(2,4-dinitrophenyl)-~-2,3-diaminopropionyl] -Ala-Arg-NHz; McaPLGLDpaAR, (7-methoxycoumarin-4-yl)acetyl-Pro-Leu-Gly-Leu- less than 200 pM (Teahan & Stein, 1990; Murphy et al., [ 3-(2,4-dinitrophenyl)-~-2,3-diaminopropionyl]-Ala-Arg-NH~. 1992a). 0006-2960/94/0433-2089$04.50/0

0 1994 American Chemical Society

2090 Biochemistry, Vol. 33, No. 8, 1994 The MMPs also have definable domains, which, in the case of gelatinase A and stromelysin- 1, include an N-terminal (catalytic) domain and a noncatalytic C-terminal domain (Matrisian, 1992). Studies aimed at understanding the nature of TIMP-MMP complex formation in the case of human gelatinaseA have identified two distinct TIMP-2 binding sites, one at the active site within the N-terminal domain and one in the noncatalytic C-terminal domain (Howard & Banda, 1991). These observations were supported by the work of Kleiner et al. (1992), who used cross-linking methods to demonstrate two binding sites for TIMP-2 on gelatinase A. Binding to one site appeared to decrease the affinity of binding of the inhibitor to the other site. Using kinetic analyses, we found that TIMP-2 bound to active gelatinase A more rapidly than TIMP-1 (Willenbrock et al., 1993). This was due to the fact that TIMP-2 has an exposed charged C-terminus which promotes rapid binding to the C-terminal domain of gelatinase A. We also found that at least three sites of interaction between TIMP-2 and gelatinase A could be defined by the study of binding rates of different inhibitor and enzyme mutants. Here, we compare the binding of human TIMP-1, TIMP-2, and deletion mutants of both TIMPs to human gelatinase A and stromelysin-1. We also compare the binding of TIMPs to deletion mutants of both enzymes lacking their C-terminal domain. By measuring the second-order rate constants for binding, we reveal that the C-terminal domain of gelatinase A contributes more to the rate of inhibitor association than the corresponding domain in stromelysin- 1. This is a unique feature of the gelatinase A C-terminal domain and is independent of the catalytic domain, since it can be transferred to stromelysin by exchanging the C-terminal domain between the two enzymes. MATERIALS AND METHODS

Chemicals. All chemicals were of the highest purity available commercially. The stromelysin substrate McaPLANvaDpaAR, (7-methoxycoumarin-4-y1)acetyl-Pro-LeuAla-Norval- [3- (2,4-dinitrophenyl)-~-2,3-diaminopropionyl] Ala-Arg-NH2, was a gift from Dr. C. G. Knight (Niedzwiecki et al., 1992; Knight et al., 1992), as was McaPLGLDpaAR (Knight et al., 1992). Preparation of Stromelysins and TIMPs. Recombinant human prostromelysin was purified from the medium of C127 cells expressing this enzyme (Murphy et al., 1987; Koklitis et al., 1991). A248460 stromelysin (N-SL) was prepared as described by Murphy et al. (1992a). Prostromelysins were activated by treatment with trypsin (10 pg/mL, 37 "C, 30 min) followed by a 10-fold excess of soybean trypsin inhibitor. TIMP-1, A127-184 TIMP-1 (N-TIMP-l), TIMP-2, A128-194 TIMP-2 (N-TIMP-2), and A187-194 TIMP-2 were prepared as described previously (Murphy et al., 1991; Willenbrock et al., 1993). Preparation of the N- Terminal Gelatinase A-C- Terminal Stromelysin Hybrid (N-GL.C-SL), the N- Terminal Stromelysin-C- Terminal Gelatinase A Hybrid (N-SL.C-GL), Progelatinase A, and A418431 Progelatinase A (N-GL). The N-terminal gelatinase A-C-terminal stromelysin-1 hybrid (gelatinase A 1-417-stromelysin 248-460, N-GL.C-SL) was assembled using the polymerase chain reaction (PCR) overlap extension method of Ho et al. (1989), pEE12.gelatinase A (Murphy et al., 1992b) was used as the template in the presence of an oligonucleotide that primes within the hCMV promoter and a second mutagenic oligonucleotide that is complementary to the sequence encoding residues 41 3-41 7 of gelatinase A and residues 248-252 of stromelysin (Figure 1) to generate

Nguyen et al.

n, N-GL.C-SL

Q E L Y G P P P D S S'-CAGGAGCTCTATGGGCCTCCCCCTGACTCC-3' 3'-GTCCTCGAGATACCCGGAGGGGGACTGAGG-5'

I

h

I

N-SL.C-GL

Q S L Y G A S P D I S'-CAGTCCCTCTATGGAGCCTCTCCTGACATT-3' 3'-GTCAGGGAGATACCTCGGAGAGGACTGTAA-5'

FIGURE 1: Construction of gelatinase A-stromelysin hybrids. The gelatin binding domain is shown as a lightly stippled region, mature gelatinase A sequences are shown in black, and mature stromelysin- 1 sequences are shown heavily stippled; the prodomain of each enzyme is shown unshaded. The putative zinc binding region within each N-terminal region is labeled, and the two cysteine residues in each C-terminus are shown. The sequences of the oligonucleotides used to generate each hybrid molecule are shown together with the amino acids encoded. Thejunctions are identified by the numbered residues of the wild-type proteins (Collier et al., 1988; Docherty & Murphy, 1990).

a 1400-bp DNA fragment. A second 780-bp fragment was generated using PEE12.stromelysin-1 (Murphy et al., 1992a) as a template, with an oligonucleotide priming within theSV40 polyadenylation sequence of PEE12 and an oligonucleotide complementary to the mutagenic oligonucleotide described above (Figure 1). The two fragments were annealed and further amplified by PCR in the presence of flanking pEE12 oligonucleotide primers. The 1500-bp HindIII fragment encoding residues 1-417 of gelatinase A and residues 248304 of stromelysin, generated by cleavage of the PCR product, was ligated into the HindIII-digested pSP65. The N-terminal ( 1-247) stromelysin-C-terminal(418-63 1) gelatinase A hybrid (N-SL.C-GL) was generated in an analogous fashion using PCR with a pair of mutagenic oligonucleotides that are complementary to the sequence encoding residues 243-247 of stromelysin- 1 and residues 4 18422 of gelatinase A (Figure 1). These generated a 900-bp 5' stromelysin-1 DNA fragment and an 830-bp 3' gelatinase A DNA fragment. The two fragments were annealed and further amplified by PCR in the presence of the flanking pEE12 primers. The 560-bp HindIII to BamHI fragment encoding residues 2 17-247 of stromelysin- 1 and residues 4 18-57 1 of gelatinase A, generated by cleavage of the polymerase chain reaction product, was ligated into pSP65 that had been digested with HindIII and BamHI. To insert the N-GL.C-SL hybrid into pEE12, a 250-bp BglI to HindIII fragment encoding residues 393-417 of gelatinase and residues 248-304 of stromelysin- 1 was purified from the vector pSP65.N-GL.C-SL and ligated in a three-way ligation to the large HindIII fragment of PEE12.stromelysin-1 containing the 3' coding sequence of stromelysin- 1 (encoding residues 305-460 in pEEl2) and a 1260-bp 5' gelatinase A HindIII to BglI fragment (encoding residues 1-392). To insert the N-SL.C-GL hybrid into pEE12, a 280-bp HindIII to BglI fragment from pSP65N-SL.C-GL encoding residues 2 17-247 of stromelysin- 1 and residues 4 18-475 of gelatinase A was ligated to HindIII, EcoRI-cut pEE12, and a 680-bp BglI to EcoRI 3' gelatinase A fragment encoding residues 476-63 1. This partial reconstruction of N-SL.C-

Biochemistry, Vol. 33, No. 8, 1994 2091

Domain Interactions of TIMPs, Stromelysin-1, and Gelatinase A GL in pEE12 was cut with HindIII and ligated with the 5' stromelysin-1 810-bp HindIII fragment from pEE12.stromelysin-1. Theconstructions PEE12.N-GL.C-SL and pEE12.N-SL.CGL were shown by dideoxy chain termination sequencing to contain the correct sequences of wild-type stromelysin- 1 or gelatinase A, as appropriate. These vectors were transfected into NSO myeloma cells by electroporation (Murphy et al., 1991), and colonies were isolated and screened for the production of metalloproteinase activities degrading casein or gelatin. The most productive cell lines were expanded and used to produce conditioned medium from which N-GL.CSL or N-SL.C-GL could be purified. N-SL.C-GL-containing medium (500 mL) was purified by passage through an S-Sepharose column after equilibration of both medium and matrix with 25 mM TrisHC1, pH 7.5, containing 10mM CaC12 and 0.02%sodium azide. The loaded column was washed with the same buffer containing 100 mM sodium chloride, and the hybrid N-SL.C-GL eluted using the same buffer containing 200 mM sodium chloride. N-GL.C-SL medium (500 mL) was purified by binding to a gelatin Sepharosecolumn equilibrated with 25 mM TrisHC1, pH 7.5, containing 500 mM NaCl, 10 mM CaC12, and 0.02% sodium azide and elution with the same buffer containing 10% (v/v) dimethyl sulfoxide. Progelatinase A and A418431 progelatinase A (N-GL) were expressed from NSO cells and purified as described previously (Murphy et al., 1992b). Activation of N-SL.C-GL hybrids was most efficiently effected with trypsin as described above for prostromelysins. Progelatinase A, N-GL, and N-GL.C-SL were activated by treatment with 2 mM (2-aminopheny1)mercuric acetate at 25 "C for 1.5 h. Proteinase Assays. Activities of the recombinant proteinases were determined using '4C-labeled @-caseinor type I gelatin, as described previously (Murphy et al., 1981). Kinetic Studies. Assays were performed at 37 "C at Z = 0.125 in a buffer containing 50 mM TrisHC1, pH 7.5, 100 mM NaC1, 10 mM CaC12, 1% dimethyl sulfoxide, and 0.05% Brij 35. Hydrolysis of the substrate McaPLANvaDpaAR or McaPLGLDpaAR was followed by the increase in fluorescence (Aex 328 nm, A,, 393 nm) using a Perkin-Elmer LS5OB fluorimeter (Knight et al., 1992; Niedzwiecki et al., 1992). Mutant TIMP concentrations were determined by titration against active stromelysin as described by Willenbrock et al. (1993). Active metalloproteinase concentrations were determined by titration against a TIMP-1 standard, the concentration of which had been determined by amino acid analysis. For this, we assumed that the preparation of TIMP-1 was 100% active and that a molar stoichiometry of binding to all the metalloproteinases and their mutants occurred. Stromelysin was generally used at a concentration of 0.30.5 nM using thesubstrate McaPLANvaDpaARandgelatinase A at a concentration of 10-100 pM using either McaPLANvaDpaAR or McaPLGLDpaAR. Substrates were used in the concentration range 0.5-1.5 pM. In each case, [SI 8 pM; Knight et al., 1992; Q . Nguyen, unpublished results), giving progress curves that were first-order in substrate and allowing direct determination of kcat/Km. Inhibition by TIMP. Reaction conditions were chosen so that in each case the TIMP concentration was greater than 10 times the enzyme concentration. Reactions were started by the addition of enzyme to solutionsof substrate and inhibitor in buffer. Data were collected until a steady-state velocity was attained, and the progress curves were analyzed by using

Table 1: Comparison of the Proteolytic Activities of the Hybrids Stromelysin (1-247)-Gelatinase A (418-631), N-SL.C-GL, and Gelatinase A (1-41 7)Stromelysin (248-460), N-GL.C-SL, with Their Wild-Type Counterparts0

gelatinase A N-GL.C-SL stromelysin-1 N-SL.C-GL

ratio casein/ gelatin-degrading activity 0.28 0.26 42.9 28.9

k,t/K, (M-'

S-')

871496 f 6370 662753 f 13100 58766 f 1160 61533 f 3164

The abilityof gelatinase A, stromelysin-1,and their hybrids todegrade 14C-labeledcasein and type I gelatin at 37 "Cwas compared. The kat/ Km values were determined at 37 OC using the quenched fluorescent peptide substrate McaPLANvaDpaAR, as described in the Materials and Methods, and are the mean of four to eight determinations. (I

the Enzfitter program (Leatherbarrow, 1987) and the equation: P = ust

+ (uo - u,)( 1 - e-k')/k

(1)

in which P is the product concentration, uo and us are the initial and steady-state velocities, respectively, for enzyme catalysis in the presence of inhibitor, and k is the apparent first-order rate constant for the establishment of equilibrium between enzyme and inhibitor complexes (Morrison and Walsh, 1988). In each case, the dependence of k on TIMP concentration was determined across as high a concentration range as was practically possible. The linear relationships observed in each case provided no evidence for a model more complex than inhibition by a simplebimolecular reaction in the concentration rangeused (see Results). Thus, thesecond-order rateconstant (ken) was provided by linear regression of k on TIMP concentration.

RESULTS Characterization of Hybrid Enzymes. In order to analyze the role of the C-terminal domains of gelatinase A and stromelysin- 1 in interactions with the TIMPs, we prepared forms of these enzymes lacking this domain (Murphy et al., 1992a,b) and hybrid enzymes consisting of the N-terminal catalytic domain of one enzyme fused to the C-terminal domain of the other (Figure 1). This is a particularly valuable approach since the C-terminal domains of all the metalloproteinases show a good degree of similarity (>50%). The N-terminal (residues 1-247) stromelysin-C terminal (residues 418-63 1) gelatinase A hybrid, N-SL-C-GL, was expressed from myeloma cells and purified using S-Sepharose. It migrated during SDS polyacrylamide gel electrophoresis (reducing conditions) as a single band with a molecular mass of 57 kDa (Figure 2B). The hybrid was apparently unglycosylated, which would be predicted from our previous observation that the potential glycosylation site of the stromelysin catalytic domain was not utilized (Murphy el al., 1992a). The hybrid could be activated by incubation with (4-aminopheny1)mercuric acetate or trypsin with similar kinetics to wild-type stromelysin-1 by loss of the 10-kDa propeptide (Figure 2B). Optimal activation, as assessed by @-caseinor McaPLANvaDpaAR assays, was obtained using trypsin at a concentration of 10 pg/mL (data not shown), which was therefore used routinely. The hybrid N-SL.C-GL degraded casein efficiently and was a poor degrader of gelatin, with a similar ratio of activities against these substrates as that of stromelysin-1 (Table 1). The N-terminal (residues

2092 Biochemistry, Vol. 33, No. 8, 1994

Nguyen et al.

A

1

2

3.4

B

97 -78

---- ----

-68

D--

91.-

e ,

* "

-55 -45 - 35

.-

-29

-9-18

1 2 3 4 5 6 7 8 9 1 0 1 1

FIGURE 2: Comparisonof the activationof gelatinaseA-stromelysin-1 hybrids with their wild-type counterparts. (A) Progelatinase A was incubated alone (lane 1) or in the presence of 2 mM (4-aminopheny1)mercuric acetate (lane 2) at 25 OC for 1 h. N-terminal gelatinaseC-terminal stromelysin,N-GL.C-SL, was incubated under the same conditions: alone (lane 3) and with (4-aminopheny1)mercuricacetate (lane 4). The products of the incubations were analyzed on an 11% polyacrylamide gel with SDS under reducing conditions and stained with Coomassie Blue. (B) N-terminal stromelysin-C-terminal gelatinase, N-SL.C-GL, was incubated at 37 OC alone for 2 h (lane 1) or with 2 mM (4-aminopheny1)mercuricacetate for 1 h (lane 2), 2 h (lane 3), and 4 h (lane 4). Treatment with trypsin (5 pg/mL) at 37 "C for 30 min followed by the addition of a 10-fold excess of soybean trypsin inhibitor elicited a similar fall in molecular mass (lane 5). Stromelysin-1 behaved similarly upon incubation under the same conditions: alone (lane 6 ) , with (4-aminopheny1)mercuric acetate for 1 h (lane 7), 2 h (lane 8), and 4 h (lane 9), or with trypsin (lane 10). Trypsin and soybean trypsin inhibitor are shown in lane 11. The incubations were analyzed on an 11% polyacrylamide gel with SDS under reducing conditions and stained with silver. The mobilities of standard molecular mass (kDa) proteins are indicated to the right of each gel.

1-417) gelatinase A-C terminal (residues 248-460) stromelysin-1 hybrid, N-GL.C-SL, could be purified on gelatin Sepharose,since it retains the fibronectin-likecollagen binding domain (Figure 1). It migrated during SDS polyacrylamide gel electrophoresis (reducing conditions) as a single band of molecular mass 72 kDa (Figure 2A). In some preparations, a minor higher molecular mass band was detectable, which may be due to glycosylation of the C-terminal domain of stromelysin-1. The hybrid could be activated by incubation with (4-aminopheny1)mercuric acetate with a similar time course to full-length gelatinase A and involving the apparent loss of the propeptide piece (Figure 2A). As in the case of wild-type gelatinase A, trypsin treatment did not cause the same shift in molecular mass (Figure 2A) nor activate the hybrid (data not shown). Optimal activation was obtained by 1-2-h treatment with the organomercurial (2 mM), as determined by gelatin or McaPLANvaDpaAR assays, and this

treatment was used routinely. The hybrid N-GL.C-SL degraded gelatin efficiently and casein poorly, with a similar ratio of activities as that of wild-type gelatinase A (Table 1). The specificitiesof stromelysin-1 and gelatinase A therefore both appear to result from the N-terminal catalytic domains of these enzymes. Inhibition Kinetics. Analyses of the inhibition of the activated purified recombinant MMPs by TIMP- 1 and TIMP-2 and their truncated forms were carried out using a continuous fluorimetric assay with the appropriate internally quenched fluorescent substrate (Neidzwiecki et al., 1992; Knight et al., 1992). Ki values for the inhibition of wild-type and mutant enzymes by the wild-type TIMPs are all less than 200 pM. In order to obtain accurate values, it is necessary to use enzyme and inhibitor concentrations which are less than the Ki. Under these conditions, long incubations (>24 h) are needed to reach a steady-state velocity, during which time enzyme activity is lost due to instability at such low concentrations. In addition, the activity assays are not sufficiently sensitive to allow reliable measurements to be performed at enzyme concentrations below the Ki value. Attempts to determine the dissociation rate constants were also prevented by the limitationsin assay sensitivity. However, half-lives for the dissociation of complexes of each enzyme and inhibitor of several hours were observed, confirming that all of the interactions are tight-binding. Our analysis was limited, therefore, to following the association of enzyme and inhibitor by using low reagent concentrations (10-50 pM gelatinase A or 0.3-0.5 nM stromelysin). Under these conditions,inhibition was observed as curvature in the progress curve of substrate hydrolysis which could be analyzed using eq 1. For each reaction, it was confirmed in the absence of inhibitor that the curvature was due to inhibition by TIMP only and not enzyme instability or substrate depletion. Analysisof the progresscurvesprovided the pseudo-first-order rate constant ( k ) for the formation of EI. The dependence of k on TIMP concentration was studied for each reaction. TIMP concentrations used were 20-400 nM for reactions with kon values of lo4 M-' s-l, 2-50 nM for kon values of lo5 M-* s-l, and 1-10 nM for kon values of lo6 M-l s-l. The reaction of gelatinaseA with TIMP-2 was studied at a TIMP concentration of 100 pM and an enzyme concentration of 10 pM. The interaction of N-SL.C-GL with full-length TIMP-1 or TIMP-2 was too rapid to be followed under pseudo-first-order conditions, and lower limits for the values of kon only were obtained. In each case, the initial rate of the enzyme-catalyzed reaction was independent of the inhibitor concentration and values of k showed a linear dependence on TIMP concentration (data not shown). Both observations provide no evidence for a mechanism involving significant concentration of intermediate states. Linear regression of k on TIMP concentration provided the value of the second-order rate constant ken. As discussed previously (Willenbrock et al., 1993), the second-order rate constant kon determined in the present work could contain a term for the interaction of TIMP with the enzyme-substrate complex as well as with the free enzyme. The results of the determination of rates for the MMPs and TIMPs are summarized in Table 2. Inhibition of full-length wild-type enzymes by wild-type TIMPs is rapid with secondorder rate constants in the range of 105-107 M-' s-l. The relative rates of inhibition vary for each enzyme; in the case of stromelysin, TIMP- 1 is a more rapid inhibitor than TIMP2. Removal of the C-terminal domain of TIMP- 1 reduces the rate constant with stromelysin by less than 2-fold, whereas

Biochemistry, Vol. 33, No. 8, I994

Domain Interactions of TIMPs, Stromelysin-1, and Gelatinase A

2093

Table 2: Rate Constants for the Interaction of TIMP-1 and -2 and Their N-Terminal Domains with Wild-Type and Mutant Forms of Gelatinase A and Stromelysin"

k,, (M-I s-I)

stromelysin A24844 stromelysin (N-SL) gelatinase A 418-631 gelatinase A (N-GL) stromelysinl-247-gelatinase41g-631(N-SL.C-GL) gelatinasel~l7-stromelysin24g~~ (N-GL.C-SL)

TIMP-1 1.9 1.9 4.7 0.014 >10 0.14

X

1W

An-184

Am-194

TIMP- 1 (N-TIMP-1) 1.17 1.2 0.17 0.14 1.58 0.15

TIMP-2 (N-TIMP-2) 0.05 0.05 0.36 0.30 0.045 0.32

TIMP-2 0.30 0.47 38 0.30 >10 0.83

The data are the mean of more than four determinations at 37 OC.

removal of the TIMP-2 C-terminal domain results in a 6-fold decrease in the association rate constant. Removal of the stromelysin C-terminal domain had little effect upon these patterns of interactions. The deletion of the nine amino acid charged "tail" from the C-terminal domain of TIMP-2 (A187-194 TIMP-2) only modified the k,, to 0.21 X lo6 M-I s-I, and variation of ionic strength did not alter the value of kon significantly (data not shown). As previously reported, inhibition of gelatinase A is characterized by a higher rate constant for TIMP-2 than for TIMP-1. Removal of the C-terminal domain of TIMP results in a dramatic decrease in the rate of inhibition: 50-fold for TIMP-1 and 100-fold for TIMP-2. Removal of the enzyme C-terminal domain also affects the rate constants markedly, with a 300-fold decrease in the rate of inhibition by TIMP-1 and a 100-fold decrease in the case of TIMP-2. Rate constants for inhibition by the truncated TIMPs, N-TIMP-1, and N-TIMP-2, are unaffected by the gelatinase A C-terminal domain. The hybrid enzymes were also inhibited rapidly by wildtype TIMPs. The hybrid consisting of the N-terminal domain of stromelysin and the C-terminal domain of gelatinase A (N-SL.C-GL) was inhibited at similar rates by TIMP-1 and TIMP-2 with k,, values considerably larger than for wildtype stromelysin. Limitations of the sensitivity of the stromelysin assay using the substrate McaPLANvaDpaAR prevented the use of enzyme concentrations lower than 0.3 nM which would permit accurate determination of the latter rate constant. The effect of removal of the TIMP C-terminal domain is to decrease the rate constants by approximately 7-fold for TIMP-1 and over 200-fold for TIMP-2. The rate of inhibition of the other hybrid, N-GL.C-SL, is relatively unaffected by the C-terminal of either TIMP, which was comparable to the behavior of wild-type stromelysin.

DISCUSSION We previously reported evidencethat the N-terminal domain of TIMP-1 is responsible for the inhibition of the matrix metalloproteinases, largely by interactions with the enzyme catalytic domain (Murphy et ai., 1991). In the case of gelatinase A, further interactions between the C-terminal domains of either TIMP-1 or TIMP-2 with the enzyme C-terminal domain play an important role in increasing the rate of formation of the inhibitory complex (Murphy et al., 1992b; Willenbrock et ai.,1993). In order to assess whether this interaction between the C-terminal domains of MMPs and TIMPs is a general phenomenon, we have now extended the study to a comparative analysis of the inhibition of stromelysin and gelatinase A. We have determined the contribution of the C-terminal domains of enzyme or inhibitor to the rate of inhibition by comparing the full-length proteins

with their genetically engineered counterparts that lack the C-terminal domain. The interaction of TIMP with each enzyme was studied by determining the rate constant for the formation of the inhibitory complex. Such an analysis does not provide information on the features of each molecule that are important in determining the stability of the complex. It does however allow us to draw some conclusions about the nature of the initial interaction. In order to do so, it is essential that the changes we have made, exchanging domains between enzymes or expressing individual catalytic domains, have not resulted in structural changes to the domains themselves. We have previously reported that the activation and catalytic activity of the N-terminal domain of both stromelysin and gelatinase A are not altered by the removal of the C-terminal domain (Murphy et al., 1992a,b). In the present work, we have also demonstrated that the properties of the N-terminal domains are not altered in the hybrid enzymes. We are confident, therefore, that differences observed between the full length, truncated, and hybrid enzymes with respect to TIMP binding are not due to conformational changes in the mutant enzymes. The conformation of the truncated form of TIMP-1 has been investigated in some detail by chemical modification studies and susceptibility to trypsin. The inhibitory activity of both forms of TIMP-1 are affected to a similar extent by modification with diethyl pyrocarbonate (Williamson et al., 1993b). Trypsin digestion not only releases the same peptides from the sequence common to both forms of TIMP but also releases them in the same order (Williamson et ai., 1993a). Therefore, it has been concluded that the structure of the N-terminal domain of TIMP- 1 is not affected by the C-terminal domain. In view of the homology between TIMP-1 and TIMP-2, it is reasonable to assume that the conformation of the TIMP-2 mutant is also correct. Each reaction studied in the present work has been analyzed for the existence of a rate-limiting step in the reaction pathway due to a conformational change in either of the reagents or the enzyme-inhibitor complex by determining the dependence of the apparent first-order rate constant k on TIMP concentration. The linear dependence observed provides no evidence for an association process more complex than a bimolecular collision. As discussed previously (Willenbrock et ai.,1993), although many of the rate constants are lower than would be expected from a diffusion-limited encounter, similar values have been reported for other enzyme interactions (see Hammes, 1982 and references cited therein). In our interpretation of the kinetic data, we therefore assume the simplest model for the interaction of a diffusion-limited bimolecular collision. This model does not preclude the possibility that conformationalchanges occur after formation of the inhibitory complex to give a final complex in which different interactions are important.

Nguyen et al.

2094 Biochemistry, Vol. 33, No. 8, 1994 In contrast to our observations on TIMP interactions with gelatinase A, the association rate constant for the TIMPstromelysin interaction is greater for TIMP-1 than for TIMP2. In addition, the rate of inhibition of stromelysin is affected to a much lesser extent by the C-terminal domain of either TIMP than is that of gelatinase A, and ionic interactions do not appear to be important in the association of the TIMPs with stromelysin. Removal of the C-terminal domain of stromelysin also does not result in any significant changes in the association rate constants, although a small increase in the kon for inhibition by TIMP-2 is observed. This differs from the results obtained with gelatinase A, in which the C-terminal domain of the enzyme is responsible for a (100300)-fold increase in the rate of inhibition by TIMP, via interactions with the TIMP C-terminal domain. Thus, the C-terminal domains of the two enzymes appear to be functionally very different with respect to TIMP binding. The replacement of the C-terminal domain of stromelysin with that of gelatinase A in the N-SL.C-GL mutant results in a dramatic increase in konvalues for both full-length TIMPs relative to those for stromelysin, whereas kon values for the truncated TIMPs are unchanged. The results obtained with the wild-type enzymes demonstrated that the TIMP C-terminal domain interacts with both the N-terminal domain of stromelysin and the C-terminal domain of gelatinase A. If the two domains (N-SL and C-GL) share a common binding site on the C-terminal domain of TIMP, then (a) binding of the C-GL domain of the hybrid enzyme would result in incorrect alignment of the enzyme and inhibitor N-terminal domains and (b) there would be competition between the domains of the hybrid enzyme for binding to TIMP. Both effects would result in a lower kon value for the inhibition of N-SL.C-GL than for the inhibition of stromelysin. The observed result of an increase in kon leads us to conclude that the C-terminal domain of TIMP must have a separate binding site for each enzyme. The results also suggest that either there is a great deal of flexibility in the enzyme and/or inhibitor which allows both sets of domains to bind efficientlyor the domain shuffling has not altered the distance between the binding sites on the two enzyme domains. The results obtained with the C-GL.N-SL hybrid differ somewhat. As for N-SL.C-GL, the reactions with the truncated TIMPs are characterized by the same rate constant as the appropriate enzyme N-terminal domain alone, confirming that the N-terminal domain of the TIMPs do not interact with the C-terminal domain of stromelysin. However, the full-length TIMPs have higher kon values than expected. In the case of the full-length TIMP-1, it is the result with truncated gelatinase A that seems to be anomalous, in that the rate constant is 10 times lower than that for N-TIMP-1. It is possible that the presence of the gelatinase C-terminal domain masks a site on the enzyme catalytic domain which interacts with the TIMP-1 C-terminal domain and is unfavorable for TIMP-1 inhibition. It is interesting to note that this site is also apparently masked by the presence of the stromelysin C-terminal domain in the hybrid. There is also an increase in kon for the inhibitionof N-GL.C-SL by TIMP-2 when compared to its interaction with N-TIMP-2 and the kon for the N-GL-TIMP-2 interaction. This suggests that there is an interaction between the C-terminal domains of stromelysin and TIMP-2, although it plays a minor role in the association of the enzyme and inhibitor. For wild-type stromelysin, this interaction actually decreases kon slightly, indicatingthat this low affinity interaction aligns the TIMP-2 inhibitory site incorrectly for the stromelysin-TIMP-2 interaction but correctly for the inhibition of N-GL.C-SL.

Stromclysin (SL)

TIMP-1

Gelatinase A (GL)

TIMP-1

NSLCGL TIMP-1

FIGURE3: Schematic representation of the binding sites involved in formation of inhibitory complexes of TIMP- 1 with stromelysin, gelatinase A, and their hybrids, N-SL.C-GL and N-GL.C-SL. The enzymes are represented as N-terminal (left side) and C-terminal domains divided by a vertical line and with different shading. TIMP-1 is denoted as consisting of two major regions made up of the N-terminal and the C-terminal three disulfide-bonded loops. The N-terminal domain contains the inhibitory site and binds the N-terminal catalytic domain of both stromelysin and gelatinase A (and their hybrids) to form a tight inactiveenzyme-inhibitor complex (site 1). The C-terminal domain of TIMP-1 also binds to theenzymes to increase the rate of inhibition of each enzyme to differing extents. There are at least two discrete binding sites within this domain, each of which interacts with different enzymes. These are depicted for stromelysin (site 2) and gelatinase A (site 3). In summary, the N-terminal domain of TIMP binds to the enzyme catalytic domain to inhibit activity (site 1 in Figure 3). This reaction can be quite slow in many cases with secondorder rate constants ranging from 5 X lo4 to 1.2 X lo6 M-' s-* for the interactions reported in the present work. The presence of the C-terminal domain of TIMP increases the rate constant for the association by binding to the N-terminal domain of stromelysin (site 2 in Figure 3) or the C-terminal domain of gelatinase A (site 3 in Figure 3). The binding sites for the two enzymes on the TIMP C-terminal domain are separate entities. The mechanism by which the binding of the TIMP C-terminal domain increases the rate of reaction need be no more complicated than increasing the probability of an interaction between the two N-terminal domains. ACKNOWLEDGMENT

We thank Mary Harrison for cell culture, Jim Turner for oligonucleotidesynthesis,and Richard Harrison and Graham Knight for the generous gifts of substrates. REFERENCES

Apte, S.S.,Mattei, M.-G., & Olsen, B. R. (1993) Genomics, in press. Boone, T. C., Johnson, M. J., De Clerck, Y.A., & Langley, K. E. (1990) Proc. Nutl. Acud. Sci. U.S.A. 87, 2800-2804. Cawston, T. E., Galloway, W. A., Mercer, E., Murphy, G., & Reynolds, J. J. (1981) Biochem. J. 195, 159-165. Cawston, T. E., Murphy, G., Mercer, E., Galloway, W. A., Hazleman, B. L., & Reynolds, J. J. (1983) Biochem. J. 211, 3 13-3 18.

Collier, I. E., Wilhelm,S.M., Eisen, A. Z., Marmer, B. L., Grant, G. A., Seltzer, J. L., Kronberger, A., He, C., Bauer, E. A., & Goldberg, G. I. (1988) J. Biol. Chem. 263, 6579-6587.

Domain Interactions of TIMPs, Stromelysin-1, and Gelatinase A DeClerck, Y .A,, Yean, T.-S., Lee, Y . ,Tokich, J. M., & Langley, K. E. (1993) Biochem. J . 289, 65-69. Docherty, A. J. P., & Murphy, G. (1990) Ann. Rheum. Dis. 49, 469-479. Docherty, A. J. P., Lyons, A., Smith, B. J., Wright, E. M., Stephens, P. E., Harris, T. J. R., Murphy, G., & Reynolds, J. J. (1985) Nature 318, 66-69. Hammes, G. G. (1982) in Enzyme Catalysis and Regulation, pp 99-102, Academic Press, New York. Ho, S. N., Hunt, H. D., Horton, R. M., Pullen, J. K., & Pease, L. R. (1989) Gene 77, 51-59. Howard, E. W., & Banda, M. (1991) J . Biol. Chem. 266,1797217917. Kleiner, D. E., Unsworth, E. J., Krutzsch, H. G., & StetlerStevenson, W. G. (1992) Biochemistry 31, 1665-1672. Knight, C. G., Willenbrock, F., & Murphy, G. (1992) FEBS Lett. 296, 263-266. Koklitis, P. A., Murphy, G., Sutton, C., & Angal, S. (1991) Biochem. J . 276, 217-221. Leatherbarrow, R. J. (1987) Enzfitter Software, Elsevier Biosoft, Cambridge, U.K. Lelievre, Y., Bouboutou, R., Boizian, J., Faucher, D., Archard, D., & Cartwright, T. (1990) Matrix 10, 292-299. Matrisian, L. M. (1992) Bioessays 14, 455-463. Morrison, J. P., & Walsh, C. T. (1988) Adu. Enzymol. Relat. Areas Mol. Biol. 61, 201-301. Murphy, G., Cawston, T. E., Galloway, W. A., Barnes, M. J., Bunning, R. A. D., Mercer, E., Reynolds, J. J., & Burgeson, R. E. (1981) Biochem. J . 199, 807-811.

Biochemistry, Vol. 33, No. 8, 1994 2095

Murphy, G., Houbrechts, A., Cockett, M. I., Williamson, R. A., O’Shea, M., & Docherty, A. J. P. (1991) Biochemistry 30, 809 7-8 1 02. Murphy, G., Allan, J. A., Willenbrock, F., Cockett, M. I., O’Connell, J. P., & Docherty, A. J. P. (1992a) J . Biol. Chem. 267, 9612-9618. Murphy, G., Willenbrock, F., Ward, R. V., Cockett, M. I., Eaton, D., & Docherty, A. J. P. (1992b) Biochem. J . 283,637-641. Niedzwiecki, L., Teahan, J., Harrison, R. K., & Stein, R. L. (1992) Biochemistry 31, 12618-12623. Okada, Y . ,Nagase, H., & Harris, E. D. (1986) J . Biol. Chem. 261, 14245-14255. O’Shea, M., Willenbrock, F., Williamson, R. A,, Cockett, M. I., Freedman, R. B., Reynolds,J. J., Docherty, A. J. P., & Murphy, G. (1992) Biochemistry 31, 10146-10152. Teahan, J. A,, & Stein, R. L. (1990) FASEB J . 4, A1977 (Abstr. 1652). Ward, R. V., Hembry, R. M., Reynolds, J. J., & Murphy, G. (1991) Biochem. J. 278, 179-187. Welgus, H. G., Jeffrey, J. J., Eisen, A. Z., Roswit, W. T., & Stricklin, G. P. (1985) Collagen Relat. Res. 5 , 167-179. Willenbrock, F., Crabbe, T., Slocombe, P. M., Sutton, C. W., Docherty, A. J. P., Cockett, M. I., O’Shea, M., Brocklehurst, K., Phillips, J. R., & Murphy, G. (1993) Biochemistry 32, 4330-4337. Williamson, R. A,, Smith, B. J., Angal, S., Murphy, G., & Freedman, R. B. (1993a) Biochim. Biophys. Acta 1164,8-16. Williamson, R. A., Smith, B. J., Angal, S., & Freedman, R. B. (1993b) Biochim. Biophys. Acta 1203, 147-154.